
import os
import numpy as np
from specvae.vae import BaseVAE
import specvae.dataset as dt
import specvae.utils as utils
import matplotlib.pyplot as plt
import seaborn as sns


# Parameters
dataset = "MoNA"
# model_name = "jointvae_20-800-200-50-3-50-200-800-20_01 (11-01-2022_23-10-10)"
# model_name = "betavae_capacity_20-800-200-50-3-50-200-800-20_01 (24-12-2021_01-50-12)"
# model_name = "betavae_capacity_100-400-100-3-400-100_01 (24-12-2021_11-06-17)"
# model_dir = "D:\\Workspace\\SpecVAE\\.model\\MoNA\\jointvae\\jointvae_20-800-200-50-3-50-200-800-20_01 (11-01-2022_23-10-10)"
# model_dir = "d:\\Workspace\\SpecVAE\\.model\\MoNA\\betavae_capacity_nextron\\betavae_capacity_20-800-200-50-3-50-200-800-20_01 (24-12-2021_01-50-12)"
# model_dir = "d:\\Workspace\\SpecVAE\\.model\\HMDB\\betavae_capacity_nextron\\betavae_capacity_20-1600-2-1600-20_02 (24-12-2021_18-27-38)"
model_dirs = [
    # MoNA
    ## BetaVAE
    ### Best
    "d:\\Workspace\\SpecVAE\\.model\\MoNA\\betavae_capacity_nextron\\best\\betavae_capacity_20-800-200-50-3-50-200-800-20_01 (24-12-2021_01-50-12)",
    "d:\\Workspace\\SpecVAE\\.model\\MoNA\\betavae_capacity_nextron\\best\\betavae_capacity_20-400-100-3-400-20_02 (24-12-2021_03-34-34)",
    "d:\\Workspace\\SpecVAE\\.model\\MoNA\\betavae_capacity_nextron\\best\\betavae_capacity_20-1600-3-1600-20_03 (24-12-2021_00-17-31)",
    "d:\\Workspace\\SpecVAE\\.model\\MoNA\\betavae_capacity_nextron\\best\\betavae_capacity_20-800-3-800-20_04 (24-12-2021_00-25-10)",
    "d:\\Workspace\\SpecVAE\\.model\\MoNA\\betavae_capacity_nextron\\best\\betavae_capacity_20-100-3-90-100-20_05 (24-12-2021_03-01-19)",
    "d:\\Workspace\\SpecVAE\\.model\\MoNA\\betavae_capacity_nextron\\best\\betavae_capacity_50-400-3-100-400-50_06 (24-12-2021_06-19-49)",
    ### Beta
    "d:\\Workspace\\SpecVAE\\.model\\MoNA\\betavae_capacity_nextron\\beta\\betavae_capacity_100-1600-200-50-3-50-200-1600-100_01 (24-12-2021_09-13-36)",
    "d:\\Workspace\\SpecVAE\\.model\\MoNA\\betavae_capacity_nextron\\beta\\betavae_capacity_100-1600-200-50-3-50-200-1600-100_02 (24-12-2021_09-15-11)",
    "d:\\Workspace\\SpecVAE\\.model\\MoNA\\betavae_capacity_nextron\\beta\\betavae_capacity_100-1600-200-50-3-50-200-1600-100_03 (24-12-2021_09-29-26)",
    "d:\\Workspace\\SpecVAE\\.model\\MoNA\\betavae_capacity_nextron\\beta\\betavae_capacity_100-1600-200-50-3-50-200-1600-100_04 (24-12-2021_09-29-36)",
    "d:\\Workspace\\SpecVAE\\.model\\MoNA\\betavae_capacity_nextron\\beta\\betavae_capacity_100-1600-200-50-3-50-200-1600-100_05 (24-12-2021_09-06-14)",
    "d:\\Workspace\\SpecVAE\\.model\\MoNA\\betavae_capacity_nextron\\beta\\betavae_capacity_100-1600-200-50-3-50-200-1600-100_06 (24-12-2021_09-05-10)",
    ### BetaVAE Score
    "d:\\Workspace\\SpecVAE\\.model\\MoNA\\betavae_capacity_nextron\\betavae_score\\betavae_capacity_100-400-100-3-400-100_01 (24-12-2021_11-06-17)",
    "d:\\Workspace\\SpecVAE\\.model\\MoNA\\betavae_capacity_nextron\\betavae_score\\betavae_capacity_100-800-100-3-800-100_02 (24-12-2021_10-59-29)",
    "d:\\Workspace\\SpecVAE\\.model\\MoNA\\betavae_capacity_nextron\\betavae_score\\betavae_capacity_100-400-200-50-3-50-200-400-100_03 (24-12-2021_09-47-19)",
    "d:\\Workspace\\SpecVAE\\.model\\MoNA\\betavae_capacity_nextron\\betavae_score\\betavae_capacity_100-1600-3-1600-100_04 (24-12-2021_08-02-44)",
    "d:\\Workspace\\SpecVAE\\.model\\MoNA\\betavae_capacity_nextron\\betavae_score\\betavae_capacity_100-400-100-3-400-100_05 (24-12-2021_10-57-01)",
    "d:\\Workspace\\SpecVAE\\.model\\MoNA\\betavae_capacity_nextron\\betavae_score\\betavae_capacity_100-400-100-3-100-400-100_06 (24-12-2021_08-41-33)",
    ### FactorVAE Score
    "d:\\Workspace\\SpecVAE\\.model\\MoNA\\betavae_capacity_nextron\\factorvae_score\\betavae_capacity_100-100-90-50-3-50-90-100-100_01 (24-12-2021_09-48-37)",
    "d:\\Workspace\\SpecVAE\\.model\\MoNA\\betavae_capacity_nextron\\factorvae_score\\betavae_capacity_100-1600-100-3-1600-100_02 (24-12-2021_10-51-41)",
    "d:\\Workspace\\SpecVAE\\.model\\MoNA\\betavae_capacity_nextron\\factorvae_score\\betavae_capacity_100-800-100-3-800-100_03 (24-12-2021_11-16-41)",
    "d:\\Workspace\\SpecVAE\\.model\\MoNA\\betavae_capacity_nextron\\factorvae_score\\betavae_capacity_100-1600-3-1600-100_04 (24-12-2021_08-02-44)",
    "d:\\Workspace\\SpecVAE\\.model\\MoNA\\betavae_capacity_nextron\\factorvae_score\\betavae_capacity_100-400-200-50-3-50-200-400-100_05 (24-12-2021_09-23-23)",
    "d:\\Workspace\\SpecVAE\\.model\\MoNA\\betavae_capacity_nextron\\factorvae_score\\betavae_capacity_100-1600-100-3-100-1600-100_06 (24-12-2021_08-25-38)",
]


device, cpu = utils.device(use_cuda=True)

GPU device count: 1
Device in use:  cuda:0


# print("Load model: %s..." % model_name)
def load_model(path, device):
    model_path = os.path.join(path, 'model.pth')
    model = BaseVAE.load(model_path, device)
    model.eval()
    return model


# labels = ['ionization_mode_id', 'collision_energy', 'total_exact_mass', 'precursor_mz', 'instrument_type_id', 'precursor_type_id', 'superclass_id', 'class_id']
if dataset == 'MoNA':
    labels = ['collision_energy', 'total_exact_mass']
    base_path = utils.get_project_path() / '.data' / 'MoNA'
    metadata_path = base_path / 'MoNA_meta.npy'
elif dataset == 'HMDB':
    labels = ['collision_energy']
    base_path = utils.get_project_path() / '.data' / 'HMDB'
    metadata_path = base_path / 'HMDB_meta.npy'

metadata = None
if os.path.exists(metadata_path):
    metadata = np.load(metadata_path, allow_pickle=True).item()


# metadata['precursor_type_id']


def load_data(target_column):
    # data_path = base_path / ('visualization_%s.csv' % target_column)
    data_path = base_path / ('%s_full.csv' % dataset)
    df = dt.Spectra.open(data_path)
    return df

def preload_data_as_tensor(df, input_columns, types, transform):
    # columns = model.config['input_columns']
    columns = input_columns
    # types = model.config['types']
    data = dt.Spectra.preload_tensor(
        device=device, data_frame=df[columns + ['id']], transform=transform, limit=-1, types=types, do_print=False)
    return data

def evaluate_model(model, df, data):
    print("Encode N=%d instances from %s dataset..." % (data['id'].shape[0], dataset))
    X, ids = data['spectrum'], data['id'] # TODO: handle the case for concatanated input
    Xrecon, z, latent_dist = model.forward_(X)
    print(z.shape)
    data_np = {}
    data_np['X'] = X.data.cpu().numpy()
    data_np['Xrecon'] = Xrecon.data.cpu().numpy()
    data_np['z'] = z.data.cpu().numpy()
    data_np['ids'] = ids
    data_np['ionization_mode_id'] = df['ionization_mode_id'].to_numpy()
    data_np['collision_energy'] = df['collision_energy'].to_numpy()
    # data_np['images'] = df['images'].to_numpy()
    return data_np


# import json
# import pandas as pd
# import plotly.express as px

# n_dim = model.latent_dim
# with open(os.path.join(model_dir, 'history.json')) as history:
#     history_logs = json.load(history)

# steps = np.array(history_logs['kldiv_cont_0'])[:,0]
# colors = ['kldiv_cont_%d' % dim for dim in range(n_dim)]
# kldivs = dict(zip(colors, [np.array(history_logs['kldiv_cont_%d' % dim])[:,1] for dim in range(n_dim)]))
# kldivs['steps'] = steps
# df = pd.DataFrame(data=kldivs)
# fig = px.line(df, x='steps', y=colors, title='KL Divergence per each cont. dimension')
# fig.show()


from IPython.display import display


df = load_data("")
df = df[
    (df['collision_energy'] == 35) & 
    (df['total_exact_mass'] <= 800) & 
    (df['instrument_type_id'] == metadata['instrument_type_id']['labels'].index('ESI-QFT')) &
    (df['ionization_mode_id'] == 1) & 
    (df['precursor_type_id'] == 2)
]
# df = df[
#     (df['collision_energy'] <= 100) & 
#     (df['total_exact_mass'] >= 248.14) & (df['total_exact_mass'] <= 249) & 
#     (df['instrument_type_id'] == metadata['instrument_type_id']['labels'].index('ESI-QFT')) & 
#     (df['ionization_mode_id'] == 1) & 
#     (df['precursor_type_id'] == 2)
# ]
display(df.groupby('total_exact_mass')['total_exact_mass'].value_counts())
display(df.groupby('collision_energy')['collision_energy'].value_counts())
display(df)

total_exact_mass  total_exact_mass
100.063663        100.063663          1
101.084064        101.084064          1
102.042927        102.042927          1
103.063329        103.063329          1
108.021129        108.021129          1
                                     ..
793.424856        793.424856          1
794.445257        794.445257          2
796.257850        796.257850          1
796.424521        796.424521          1
798.221858        798.221858          1
Name: total_exact_mass, Length: 1299, dtype: int64

collision_energy  collision_energy
35.0              35.0                2020
Name: collision_energy, dtype: int64


from scipy import stats
from specvae.jointvae import JointVAE
from specvae.vae import SpecVEA
# cc = model.config['cont_capacity']
# cols, vals = ['model_name', 'min_capacity', 'max_capacity', 'num_iter', 'gamma_z'], [model_name] + cc

from sklearn.metrics import r2_score, mean_squared_error, explained_variance_score
from sklearn.linear_model import LinearRegression


for model_path in model_dirs:
    print("-------------------------------------------")
    print(model_path)

    model = load_model(model_path, device)
    data = preload_data_as_tensor(df, input_columns=model.config['input_columns'], types=model.config['types'], transform=model.transform)
    data_np = evaluate_model(model, df, data)

    for target_column in labels:
        print(target_column)

        if isinstance(model, SpecVEA):
            n_dim = data_np['z'].shape[1]
        elif isinstance(model, JointVAE):
            n_dim = model.config['latent_spec']['cont']

        def plot_reg(df, ax, data_np, dim=0):
            df['z'] = data_np['z'][:,dim]
            slope, intercept, r_value, p_value, std_err = stats.linregress(df[target_column], df['z'])

            x, y = df[target_column].to_numpy().reshape(-1,1), df['z'].to_numpy().reshape(-1,1)
            reg = LinearRegression().fit(x, y)
            y_ = reg.predict(x)
            r2_value = r2_score(y, y_)
            rmse_value = mean_squared_error(y, y_, squared=False)
            exp_variance_value = explained_variance_score(y, y_)
            pearson_value, pp = stats.pearsonr(df[target_column], df['z'])
            spearman_value, sp = stats.spearmanr(df[target_column], df['z'])

            print("r^2", r2_value)
            print("RMSE", rmse_value)
            print("explained variance", exp_variance_value)
            print("Pearson", pearson_value, "p=", pp)
            print("Spearman", spearman_value, "p=", sp)
            print("-----------------")
            
            if target_column == 'collision_energy':
                sns.boxplot(x=target_column, y='z', data=df, ax=ax)
            else:
                sns.regplot(x=target_column, y='z',
                    data=df, color='blue', scatter_kws={'s': 5}, ax=ax,
                    line_kws={'label':"z[{0}] = {1:.2f} + {2:.3f}*10^-3 * {3}".format(dim, intercept, slope * 1000., target_column)})
                ax.legend(fontsize=16)
        
        fig, axs = plt.subplots(1, n_dim, figsize=(15 * n_dim, 10))
        if n_dim == 1:
            plot_reg(df, axs, data_np, 0)
        else:
            for dim, ax in enumerate(axs):
                plot_reg(df, ax, data_np, dim)

        sns.despine(right = True)
        plt.show()
        plt.clf()

-------------------------------------------
d:\Workspace\SpecVAE\.model\MoNA\betavae_capacity_nextron\best\betavae_capacity_20-800-200-50-3-50-200-800-20_01 (24-12-2021_01-50-12)
Encode N=2020 instances from MoNA dataset...
torch.Size([2020, 3])
collision_energy
r^2 -4.147547416621933e-09
RMSE 0.664303077446825
explained variance -1.0188360821672404e-08
Pearson nan p= nan
Spearman nan p= nan
-----------------
r^2 1.1425840718004565e-08
RMSE 0.7824517928029603
explained variance 4.7832540928105516e-08
Pearson nan p= nan
Spearman nan p= nan
-----------------
r^2 -2.584761493729104e-08
RMSE 0.5377247045512953
explained variance -1.39494702455778e-08
Pearson nan p= nan
Spearman nan p= nan
-----------------

d:\Workspace\anaconda3\envs\specvae\lib\site-packages\scipy\stats\_stats_mstats_common.py:170: RuntimeWarning: invalid value encountered in double_scalars
  slope = ssxym / ssxm
d:\Workspace\anaconda3\envs\specvae\lib\site-packages\scipy\stats\_stats_mstats_common.py:187: RuntimeWarning: divide by zero encountered in double_scalars
  slope_stderr = np.sqrt((1 - r**2) * ssym / ssxm / df)
d:\Workspace\anaconda3\envs\specvae\lib\site-packages\scipy\stats\stats.py:4023: PearsonRConstantInputWarning: An input array is constant; the correlation coefficient is not defined.
  warnings.warn(PearsonRConstantInputWarning())
d:\Workspace\anaconda3\envs\specvae\lib\site-packages\scipy\stats\stats.py:4484: SpearmanRConstantInputWarning: An input array is constant; the correlation coefficient is not defined.
  warnings.warn(SpearmanRConstantInputWarning())
d:\Workspace\anaconda3\envs\specvae\lib\site-packages\scipy\stats\_stats_mstats_common.py:170: RuntimeWarning: invalid value encountered in double_scalars
  slope = ssxym / ssxm
d:\Workspace\anaconda3\envs\specvae\lib\site-packages\scipy\stats\_stats_mstats_common.py:187: RuntimeWarning: divide by zero encountered in double_scalars
  slope_stderr = np.sqrt((1 - r**2) * ssym / ssxm / df)
d:\Workspace\anaconda3\envs\specvae\lib\site-packages\scipy\stats\stats.py:4023: PearsonRConstantInputWarning: An input array is constant; the correlation coefficient is not defined.
  warnings.warn(PearsonRConstantInputWarning())
d:\Workspace\anaconda3\envs\specvae\lib\site-packages\scipy\stats\stats.py:4484: SpearmanRConstantInputWarning: An input array is constant; the correlation coefficient is not defined.
  warnings.warn(SpearmanRConstantInputWarning())
d:\Workspace\anaconda3\envs\specvae\lib\site-packages\scipy\stats\_stats_mstats_common.py:170: RuntimeWarning: invalid value encountered in double_scalars
  slope = ssxym / ssxm
d:\Workspace\anaconda3\envs\specvae\lib\site-packages\scipy\stats\_stats_mstats_common.py:187: RuntimeWarning: divide by zero encountered in double_scalars
  slope_stderr = np.sqrt((1 - r**2) * ssym / ssxm / df)
d:\Workspace\anaconda3\envs\specvae\lib\site-packages\scipy\stats\stats.py:4023: PearsonRConstantInputWarning: An input array is constant; the correlation coefficient is not defined.
  warnings.warn(PearsonRConstantInputWarning())
d:\Workspace\anaconda3\envs\specvae\lib\site-packages\scipy\stats\stats.py:4484: SpearmanRConstantInputWarning: An input array is constant; the correlation coefficient is not defined.
  warnings.warn(SpearmanRConstantInputWarning())

total_exact_mass
r^2 0.04287237338309713
RMSE 0.649906962198283
explained variance 0.04287236760126767
Pearson 0.20705645933616637 p= 5.336316971653101e-21
Spearman 0.23250008050801355 p= 3.3682239599386913e-26
-----------------
r^2 0.006660683118936417
RMSE 0.7798416118681303
explained variance 0.00666071928314349
Pearson -0.08161293873644945 p= 0.00024075255569202278
Spearman -0.07078925190482395 p= 0.001454527588149293
-----------------
r^2 0.04715755133557631
RMSE 0.5248926993316515
explained variance 0.04715756267263316
Pearson 0.2171579516487489 p= 5.504075455824221e-23
Spearman 0.24839934157367277 p= 8.766254352946442e-30
-----------------

<Figure size 432x288 with 0 Axes>

-------------------------------------------
d:\Workspace\SpecVAE\.model\MoNA\betavae_capacity_nextron\best\betavae_capacity_20-400-100-3-400-20_02 (24-12-2021_03-34-34)
Encode N=2020 instances from MoNA dataset...
torch.Size([2020, 3])
collision_energy
r^2 -3.6947340742443657e-09
RMSE 0.575849561488775
explained variance -7.4977525965636e-08
Pearson nan p= nan
Spearman nan p= nan
-----------------
r^2 4.1068646261521735e-09
RMSE 0.4373118601582907
explained variance 1.9030944597986377e-08
Pearson nan p= nan
Spearman nan p= nan
-----------------
r^2 -2.7252179624071005e-08
RMSE 0.7190397377663857
explained variance 1.0537313255287728e-07
Pearson nan p= nan
Spearman nan p= nan
-----------------

d:\Workspace\anaconda3\envs\specvae\lib\site-packages\scipy\stats\_stats_mstats_common.py:170: RuntimeWarning: invalid value encountered in double_scalars
  slope = ssxym / ssxm
d:\Workspace\anaconda3\envs\specvae\lib\site-packages\scipy\stats\_stats_mstats_common.py:187: RuntimeWarning: divide by zero encountered in double_scalars
  slope_stderr = np.sqrt((1 - r**2) * ssym / ssxm / df)
d:\Workspace\anaconda3\envs\specvae\lib\site-packages\scipy\stats\stats.py:4023: PearsonRConstantInputWarning: An input array is constant; the correlation coefficient is not defined.
  warnings.warn(PearsonRConstantInputWarning())
d:\Workspace\anaconda3\envs\specvae\lib\site-packages\scipy\stats\stats.py:4484: SpearmanRConstantInputWarning: An input array is constant; the correlation coefficient is not defined.
  warnings.warn(SpearmanRConstantInputWarning())
d:\Workspace\anaconda3\envs\specvae\lib\site-packages\scipy\stats\_stats_mstats_common.py:170: RuntimeWarning: invalid value encountered in double_scalars
  slope = ssxym / ssxm
d:\Workspace\anaconda3\envs\specvae\lib\site-packages\scipy\stats\_stats_mstats_common.py:187: RuntimeWarning: divide by zero encountered in double_scalars
  slope_stderr = np.sqrt((1 - r**2) * ssym / ssxm / df)
d:\Workspace\anaconda3\envs\specvae\lib\site-packages\scipy\stats\stats.py:4023: PearsonRConstantInputWarning: An input array is constant; the correlation coefficient is not defined.
  warnings.warn(PearsonRConstantInputWarning())
d:\Workspace\anaconda3\envs\specvae\lib\site-packages\scipy\stats\stats.py:4484: SpearmanRConstantInputWarning: An input array is constant; the correlation coefficient is not defined.
  warnings.warn(SpearmanRConstantInputWarning())
d:\Workspace\anaconda3\envs\specvae\lib\site-packages\scipy\stats\_stats_mstats_common.py:170: RuntimeWarning: invalid value encountered in double_scalars
  slope = ssxym / ssxm
d:\Workspace\anaconda3\envs\specvae\lib\site-packages\scipy\stats\_stats_mstats_common.py:187: RuntimeWarning: divide by zero encountered in double_scalars
  slope_stderr = np.sqrt((1 - r**2) * ssym / ssxm / df)
d:\Workspace\anaconda3\envs\specvae\lib\site-packages\scipy\stats\stats.py:4023: PearsonRConstantInputWarning: An input array is constant; the correlation coefficient is not defined.
  warnings.warn(PearsonRConstantInputWarning())
d:\Workspace\anaconda3\envs\specvae\lib\site-packages\scipy\stats\stats.py:4484: SpearmanRConstantInputWarning: An input array is constant; the correlation coefficient is not defined.
  warnings.warn(SpearmanRConstantInputWarning())

<Figure size 432x288 with 0 Axes>

total_exact_mass
r^2 0.06311685877862572
RMSE 0.5573804754653673
explained variance 0.06311679199497988
Pearson -0.25123069525867997 p= 1.8915136167960563e-30
Spearman -0.2774695072394187 p= 4.980273392544591e-37
-----------------
r^2 0.06681632977168328
RMSE 0.4224495209425519
explained variance 0.06681634369859091
Pearson -0.2584885412145461 p= 3.39522257607939e-32
Spearman -0.2732499067394259 p= 6.392618278599651e-36
-----------------
r^2 0.0017724804542273986
RMSE 0.7184022034116205
explained variance 0.0017726128444601752
Pearson 0.042101159818969565 p= 0.05850705806282815
Spearman 0.059668841438603265 p= 0.0073071714544657275
-----------------

<Figure size 432x288 with 0 Axes>

-------------------------------------------
d:\Workspace\SpecVAE\.model\MoNA\betavae_capacity_nextron\best\betavae_capacity_20-1600-3-1600-20_03 (24-12-2021_00-17-31)
Encode N=2020 instances from MoNA dataset...
torch.Size([2020, 3])
collision_energy
r^2 -5.0956425745596334e-09
RMSE 0.1810380104357133
explained variance 2.948580490880204e-09
Pearson nan p= nan
Spearman nan p= nan
-----------------
r^2 -5.004691994159316e-09
RMSE 0.3394804898056683
explained variance 6.728290591340169e-09
Pearson nan p= nan
Spearman nan p= nan
-----------------
r^2 -2.3316085062674574e-08
RMSE 0.708820351174095
explained variance 7.17245546288936e-08
Pearson nan p= nan
Spearman nan p= nan
-----------------


# df2 = pd.DataFrame([vals], columns=cols)
# df2


# stats_file = "d:\\Workspace\\SpecVAE\\.model\\MoNA\\joint_vae_%d_stats.csv" % n_dim
# if os.path.exists(stats_file):
#     df = pd.read_csv(stats_file, index_col=0)
#     df = pd.concat([df, df2], ignore_index=True)
#     df.to_csv(stats_file)
# else:
#     df2.to_csv(stats_file)

	Unnamed: 0	spectrum	SMILES	instrument	library	author	publication	structural_key	CASMI	split	...	instrument_type_id	precursor_type_id	kingdom	superclass	class	subclass	kingdom_id	superclass_id	class_id	subclass_id
6	6	50.764058:0.030004 50.861465:0.032838 50.99858...	O=C1OC=2C=C3OC(CC3=CC2C=C1C(C=C)(C)C)C(OC(=O)C...	Thermo Q Exactive HF	Vaniya/Fiehn Natural Products Library	Arpana Vaniya	NaN	AWMHMGFGCLBSAY	NaN	train	...	0	2	Organic compounds	Phenylpropanoids and polyketides	Coumarins and derivatives	Furanocoumarins	1	18	70	184
39	39	50.093486:0.010973 54.760810:0.010155 54.85744...	OC1CC(=C)C2CC(CCC2(C)C1)C(O)(C)C	Thermo Q Exactive HF	Vaniya/Fiehn Natural Products Library	Arpana Vaniya	NaN	XZXBGGYJQALVAW	NaN	train	...	0	2	Organic compounds	Lipids and lipid-like molecules	Prenol lipids	Sesquiterpenoids	1	6	198	420
56	56	51.377261:0.015250 51.667789:0.017698 52.09552...	O=C(OC1OC(COC(=O)C(=C)C(O)CO)C(O)C(O)C1O)C=CC=...	Thermo Q Exactive HF	Vaniya/Fiehn Natural Products Library	Arpana Vaniya	NaN	MYWZGSROPRXVPM	NaN	train	...	0	2	Organic compounds	Lipids and lipid-like molecules	Saccharolipids	NaN	1	6	233	-1
80	80	50.130802:0.007823 50.143618:0.006986 50.25242...	O=C(O)C1=COC(OC2OC(CO)C(O)C(O)C2O)C3C(=CCC13)C...	Thermo Q Exactive HF	Vaniya/Fiehn Natural Products Library	Arpana Vaniya	NaN	VTQUQEWGIJRVHB	NaN	train	...	0	2	Organic compounds	Lipids and lipid-like molecules	Prenol lipids	Terpene glycosides	1	6	198	446
106	106	50.303740:0.030832 50.373785:0.030336 50.53019...	O=C1OC=C(C=C1)C2CCC3(O)C4CCC5=CC(OC6OC(CO)C(OC...	Thermo Q Exactive HF	Vaniya/Fiehn Natural Products Library	Arpana Vaniya	NaN	KKOLMSSIDZXPLS	NaN	train	...	0	2	Organic compounds	Lipids and lipid-like molecules	Steroids and steroid derivatives	Steroid lactones	1	6	238	429
...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...
120095	120095	50.147354:0.008438 51.209227:0.008929 52.42235...	O=C1C=2C=CC=CC2OC3=CC=C(OC)C=C31	Thermo Q Exactive HF	Vaniya/Fiehn Natural Products Library	Arpana Vaniya	NaN	DVZCOQQFPCMIPO	NaN	train	...	0	2	Organic compounds	Organoheterocyclic compounds	Benzopyrans	1-benzopyrans	1	14	43	11
120101	120101	51.215130:0.007352 51.361517:0.007893 52.89855...	C=1C=CC2=C(C1)NC(=C2CN(C)C)C	Thermo Q Exactive HF	Vaniya/Fiehn Natural Products Library	Arpana Vaniya	NaN	AJNMXMZCRHTBEH	NaN	valid	...	0	2	Organic compounds	Organoheterocyclic compounds	Indoles and derivatives	Indoles	1	14	124	237
120551	120551	50.057374:0.014687 50.554329:0.015111 50.80970...	O=C1OC2C(C1=C)CCC3(C)C(O)CCC(=C)C23	Thermo Q Exactive HF	Vaniya/Fiehn Natural Products Library	Arpana Vaniya	NaN	FKBUODICGDOIGB	NaN	train	...	0	2	Organic compounds	Lipids and lipid-like molecules	Prenol lipids	Terpene lactones	1	6	198	447
124855	124855	51.399748:0.018784 51.908693:0.024215 51.93020...	O1C2=CC=C(C=C2OC1)C3OCC4C(OCC34)C5=CC=C6OCOC6=C5	Thermo Q Exactive HF	Vaniya/Fiehn Natural Products Library	Arpana Vaniya	NaN	PEYUIKBAABKQKQ	NaN	train	...	0	2	Organic compounds	Lignans, neolignans and related compounds	Furanoid lignans	NaN	1	5	99	-1
125729	125729	51.202819:0.069772 53.385390:0.050123 54.10081...	O=C(OC1CCC2(C)C(CCC3(C)C2CC=C4C5C(C)C(C)CCC5(C...	Thermo Q Exactive HF	Vaniya/Fiehn Natural Products Library	Arpana Vaniya	NaN	HLBZSQOUBVLLLL	NaN	train	...	0	2	Organic compounds	Lipids and lipid-like molecules	Prenol lipids	Triterpenoids	1	6	198	467

Swarm analysis¶

Load model¶

Load and transform data¶

Load training logs¶